Since the tumultuous events of 2007, much work has suggested that financial shocks are the main driver of economic fluctuations. In a recent paper, I propose a novel strategy to identify financial disturbances. I use the evolution of loan finance relative to bond finance to proxy for firms’ credit conditions and single out the shocks born in the financial sector. I apply and test the method for the US economy. I obtain three key results. First, financial shocks account for around a third of the US business cycle. Second, these shocks occur around precise events such as the Japanese crisis and the Great Recession. Third, the financial shocks I obtain are predictive of the corporate bond spread.
Looking for the financial shock
What makes it so difficult to isolate economic fluctuations born in the financial sector? Why not just use the usual credit spreads and asset prices to proxy firms’ credit conditions?
The reason is simple. Because virtually all shocks propagate via credit conditions, this makes credit spreads and asset prices responsive to pretty much all economic and non-economic events and, as such, quite arduous to interpret. Another muddling factor to be considered is the difficulty to observe credit conditions (see for instance Romer and Romer (2017)). Raising debt requires borrowers’ compliance with countless binding agreements. If credit costs decrease while loan covenants tighten up, have credit conditions eased or not?
I use an off-the-shelf dynamic stochastic general equilibrium (DSGE) model to illustrate these points. I show that under general conditions, shocks shifting firms’ credit conditions through second-round effects are indistinguishable from shocks that directly impinge on credit conditions. Following Uhlig and De Fiore (2011), I then extend the model so that firms can fund production using either loans or bonds. Loan funding is more expensive but allows for renegotiation depending on borrowers’ productivity.
This new framework has a decisive implication: because a sudden change in credit conditions drives a wedge between the costs of bonds and loans, financial shocks are now the only type of disturbances causing opposite movements in the volumes of the two types of funding.
Let the data speak
Based on this finding from the DSGE model, I use sign-restriction techniques within a simple VAR model to capture financial shocks and identify the sources of US economic fluctuations. More specifically, I identify financial shocks as the only type of disturbances that entail opposite movements in loan and bond volumes. The VAR model allows me to study the responses in investment, prices, and the policy rate to financial shocks. Despite imposing only a minimal set of restrictions on financial shocks, I find they imply impulse responses in line with predictions from various DSGE models for all variables.
Figure 1: Historical shock decomposition for US GDP
Note: Contribution of the different structural shocks to output fluctuations. Grey areas correspond to NBER recession dates.
Figure 1 displays the historical shock decomposition for US GDP between 1985 and 2018. Financial shocks weigh on output growth during the Japanese banking crisis of the early 90s’ and the Russian crisis of 1998. While they constitute one of the leading forces driving US fluctuations, other disturbances on the demand and supply sides of the economy play a significant role, especially during the Great Recession.
Putting the model to the test
The mechanism of debt arbitrage, firms shifting between bond and loan finance, in the DSGE model is the foundation of our identification strategy. Can we provide some tests for it? The answer is yes. I take advantage of the structural nature of the DSGE model to construct the series of financial disturbances that maximizes the model’s likelihood for various US series over the period 1985 to 2018. I study its implications for credit costs, a series I have ignored so far.
Figure 2 shows our index together with the bond spread for US non-financial corporates. The two series are highly correlated. More importantly, the credit shocks captured within the DSGE model are predictive of the bond spread.
Figure 2: Financial stress and the bond spread
Note: The orange line corresponds to the estimate of the financial shock in the DSGE model. The blue line corresponds to the Moody’s seasoned AAA corporate bond rate minus the federal funds rate. Grey areas correspond to NBER recession dates.
This simple exercise highlights two essential properties for our identification strategy. First, the method yields financial shocks that are related to credit spreads, a necessary condition for any operational measure of financial stress. Second, it suggests an explanation for shifts in borrowing costs based on changes in firms’ debt financing following shocks to credit conditions.
In a nutshell
Identifying financial disturbances is difficult as financial variables can respond to all sorts of events. I propose to bypass this endogeneity issue by using firms’ debt arbitrage to identify financial shocks. I find these shocks account for a large share of the business cycle. The method highlights the importance of firms’ debt arbitrage, both as a gauge for credit conditions and as an explanation for shifts in credit spreads. Some practical advantages of the approach are worth mentioning. First, the method is easy to implement and produces results in line with more involved techniques. Second, it is model-based: the conditions for the validity of the identification scheme are set out clearly and can be modified to investigate its robustness.
David Gauthier works in the Bank’s Research Hub Secondees Division.
If you want to get in touch, please email us at email@example.com or leave a comment below.
Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge — or support — prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.